Cancer-associated fibroblasts promote osimertinib resistance in non-small cell lung cancer cells via METTL1-mediated NET1 mG modification.

Introduction
Lung cancer is the most lethal cancer worldwide, and NSCLC accounts for approximately 85% of total lung cancer case. Epidermal growth factor receptor (EGFR) is one of the key driver oncogenes in NSCLC. Clinical studies have demonstrated that the EGFR mutation-targeted therapy with tyrosine kinase inhibitors (TKIs) has achieved significantly survival benefit compared to standard platinum-based chemotherapy. Osimertinib is a third-generation irreversible TKI, which inhibits both EGFR-TKI-sensitizing and EGFR T790M resistance mutations. Although osimertinib treatment has obtained longer overall survival than the first-generation EGFR-TKIs such as gefitinib and erlotinib [1], the acquired resistance to osimertinib is inevitable. The resistance mechanism to osimertinib is complicated, and can be classified into two main categories: the EGFR-dependent mechanism including EGFR mutations such as C797X; and the EGFR-independent mechanism, including gene amplifications, gene mutations, signaling pathway activation, phenotype transformation [2], however, the mechanism is still largely unknown, and osimertinib resistance remains a major challenge in the treatment of lung cancer.
It is well established that TME plays a crucial role in tumorigenesis and tumor progression. CAFs are major components of TME, they are involved in tumor development, in particular, in drug resistance. CAFs-derived COL17A1 promotes gemcitabine resistance in pancreatic cancer cells through interacting with ACTN4 [3]. CAFs cause doxorubicin resistance of triple-negative breast cancer via histone lactylation-mediated ZFP64 up-regulation and ferroptosis suppression [4]. Our previous study has demonstrated that CAFs contribute to cisplatin resistance in lung cancer cells via secreting ANXA3 and activating JNK pathway [5]. Recent studies have indicated that CAFs are also involved in TKIs resistance. CAFs-derived phosphoprotein 1 enhances resistance of hepatocellular carcinoma to sorafenib and lenvatinib via activation of RAF/MAPK and PI3K/AKT pathways, and epithelial-mesenchymal transition (EMT) process [6]. CAFs promote almonertinib resistance in NSCLC through HK2-mediated glycolysis and SKP2 signaling [7]. Interestingly, Wei G et al. reported that almonertinib treatment promoted the accumulation of CAFs in NSCLC, and accumulation of CAFs stimulated almonertinib resistance via activation of the YAP/TAZ signaling pathway [8]. There are limited studies showed that CAFs also promote osimertinib resistance in NSCLC cells [9, 10], however, whether m7G modification is involved or not has not been elucidated.
RNA modification is one of the key epigenetic modifications, it regulates biological function of RNA at the post-transcriptional level. RNA modification has been implicated in diverse physiological and pathological processes, in particular, in cancer development [11]. Recently, we investigated the role of RNA m6A modification in NSCLC metastasis. We have revealed that CAFs facilitate metastasis of NSCLC cells via METTL3-mediated RAC3 m6A modification, we also found that phenethyl isothiocyanate suppresses metastasis potential of NSCLC cells through FTO mediated TLE1 m6A modification [12, 13]. Besides m6A modification, m7G modification is also one of the most abundant RNA modifications. m7G modification is involved in synthesis and processing of messenger RNA (mRNA), ribosomal RNA (rRNA), transfer RNA (tRNA), and microRNA (miRNA). In mRNA, m7G modification is primarily catalyzed by METTL1 and WD repeats domain 4 (WDR4) complex [11]. Numerous studies have demonstrated that dysregulated m7G modification promotes tumor progression via modulating cell proliferation, migration and invasion, energy metabolism, and immune evasion. Notably, aberrant m7G modification is associated with tumor staging, lymph node metastasis, and poor prognosis [14]. However, the role of m7G modification in the resistance of NSCLC to osimertinib has not been evaluated.
In this study, we investigated the role of CAFs in the resistance of NSCLC cells to osimertinib, and explored the underling mechanism, in particular, the role of m7G modification in NSCLC cells regulated by CAFs. We further identified the m7G modified gene and its downstream signaling pathway. Our study will provide useful information for finding novel therapeutic strategies to overcome osimertinib resistance of NSCLC by targeting m7G modification.

Materials and methods

Reagents and antibodies
Osimertinib was purchased from MCE (Shanghai, China). Recombinant human HMGB1 was purchased from Chimerigen Laboratories (Allston, MA). Antibodies against METTL1 (#14994-1-AP), NET1 (#28180-1-AP), OCT4 (#11263-1-AP) (1:1000 dilution) were purchased from Proteintech Group (Rosemont, IL). Antibodies against Survivin (#2808), Caspase-3 (#9662), Cleaved-caspase-3 (#9661), AKT (#4691), p-AKT (#4060), p65 (#4764), p-p65 (#3033) (1:1000 dilution) were purchased from Cell Signaling Technology (Beverly, MA). Antibodies against SOX2 (#A11501), Nanog (# A22625) (1:1000 dilution) were purchased from ABclonal (Wuhan, China). Anti-m7G (#4141-131, 1:1000 dilution) was purchased from MBL Beijing Biotech Co. (Beijing, China); anti-β-actin (#A2228, 1:4000 dilution) was purchased from Sigma-Aldrich (St. Louis, MO); goat anti-rabbit IgG (#ZB-2301, 1:8000 dilution) and anti-mouse IgG (#ZB2305, 1:8000 dilution) were purchased from ZSGB- BIO (Beijing, China). HMGB1 neutralizing antibody was purchased from Abnova (Taipei, Taiwan Province, China).

Cell culture
NSCLC cell lines PC9 and H1975 were purchased from National Collection of Authenticated Cell Cultures (Shanghai, China). BEAS-2B cells was purchased from the American Type Culture Collection (Manassas, VA). All cell lines were confirmed via STR profiling. PC9 cells and BEAS-2B cells were cultured in DMEM medium, and H1975 cells were cultured in RPMI-1640 medium, supplied with 10% fetal bovine serum (GIBCO; Grand Island, NY). Fibroblasts were isolated from cancer and adjacent noncancerous tissues of NSCLC patients who underwent surgery at Tianjin Medical University General Hospital (TMUGH, Tianjin, China), and α-SMA and FAP were used for CAFs identification. The CAF conditioned medium (CAF-CM) was collected as previously described, CAFs and NFs were less than 10 passages [15].

Cell proliferation assay
Cells were cultured in 96-well plates at a density of 5 × 103 cells/well. Cells were treated accordingly and continued to grow for 48 h. The CCK-8 kit (Dojindo, Kumamoto, Japan) was used to assess cell proliferation according to the manufacturer’s instructions. The median inhibitory concentration IC50 values were calculated by using GraphPad Prism 9 (GraphPad Software, San Diego, CA).

Colony formation assay
Cells were cultured in 6-well plates at a density of 1000 cells/well. Cells grew continuously for 1-2 weeks. After the formation of cell colonies, colonies were stained with crystal violet, images were collected, and colonies were counted.

Apoptosis assay
Cells were cultured in 6-well plates at a density of 2 × 105 cells/well. Cells were treated accordingly and continued to grow for 48 h. Apoptosis was detected using an AnnexinV/PI Apoptosis Detection Kit (Solarbio, Beijing, China) according to the manufacturer’s instructions, and quantified on a flow cytometer (Becton Dickenson, San Jose, CA).

Enzyme-linked immunosorbent assay (ELISA)
Cell culture supernatants were collected, and HMGB1 levels in medium were measured using an ELISA kit as previously described [16].

RNA interference and plasmid transfection
siMETTL1, siNET1, siRNA control (Tsingke Biotechnology, Beijing, China), or pMETTL1, pNET1, and control vector (Tsingke Biotechnology) were transfected into cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). The sequences of siRNA duplex were as follows: for METTL1, sense 5’- GGAAGAAAGUUCUACGUAATT -3’, antisense 5’- UUACGUAGAACUUUCUUCCTT -3’; for NET1, sense 5’- CAGAAUCGAAGCGAGCAAA -3’, antisense 5’- UUUGCUCGCUUCGAUUCUG -3’; for control, sense 5’- UUCUCCGAACGUGUCACGUTT -3’, antisense 5’- ACGUGACACGUUCGGAGAATT -3’.

Western blotting
Cells were lysed using the RIPA buffer (Beyotime, Shanghai, China) containing protease inhibitors (Sigma-Aldrich). Protein concentration was measured using a BCA protein assay kit (Beyotime). Proteins were separated by 12% SDS-PAGE and transferred onto a nitrocellulose membrane (Millipore, St. Louis, MO). After 1 h of blocking with 5% milk, the membranes were probed with primary antibodies and the corresponding HRP-conjugated secondary antibodies. Finally, the protein bands were visualized using the ECL system (Thermo Fisher Scientific, Waltham, MA).

Quantitative PCR (qPCR)
RNA was extracted using TRIzol (Invitrogen, Carlsbad, CA), and reverse transcriptions were performed using the Takara kit (Dalian, China). Gene expressions were assessed by qPCR using the Power SYBR-Green Master Mix (Thermo Fisher Scientific). PCR primer sequences were shown in Table 1. GAPDH was used as an internal control.

RNA m7G dot blot
Dot blots were performed as previously described [13]. Briefly, total RNA was extracted using TRIzol (Invitrogen). 250 ng of RNA was loaded on a nylon membrane (Beyotime), and crosslinking of RNA to the nylon membrane was performed in an UV crosslinker. The membrane was blocked with 5% milk then incubated with anti-m7G antibody. After washing the membranes were probed with the corresponding secondary antibodies and the spots were visualized using the ECL system (Thermo Fisher Scientific). Another nylon membrane loaded with same amount of RNA were stained with 0.02% methylene blue (Sangon Biotech, Shanghai, China), used for loading control.

Methylated RNA immunoprecipitation sequencing (MeRIP-seq)
MeRIP-seq was performed as previously described [12]. Briefly, RNA was extracted and sheared into fragments of approximately 100 nt. The RNA fragments were incubated with anti-m7G antibody coupled-protein A/G magnetic beads (Thermo Fisher Scientific) overnight at 4°C. Then the antibody-bound methylated RNA was used to performed MeRIP-seq by Cloudseq (Shanghai, China), and the RNA-seq was performed parallelly.
Cutadapt software (v1.9.3) was used to remove adapt information and remove low quality reads to obtain high quality clean reads. Methylated genes in each sample were then identified using MACS software (v1.4.2). Differentially methylated gene identification was performed using diffReps software (v1.55.6). The peaks located on mRNA were screened and annotated accordingly. GO and KEGG analyses were performed by clusterProfiler (v4.12.6) for differentially methylated mRNA. False discovery rate (FDR) was calculated with egdeR software (v4.0.16).

Methylated RNA immunoprecipitation quantitative PCR (MeRIP-qPCR)
RNA was extracted and purified, and 1/10 of total RNA was used as an input control. Briefly, protein A/G magnetic beads were incubated with anti-m7G antibody for 2 h at 4°C, then the m7G antibody-coupled protein A/G magnetic beads complexes were mixed with RNA in immunoprecipitation buffer. The methylated mRNA was precipitated and the enrichment of target gene was determined by qPCR.

RNA immunoprecipitation (RIP)
RIP assay was performed as previously described [12], using the Magna RIP RNA-binding protein immunoprecipitation kit (Millipore, St. Louis, MO). Briefly, METTL1 antibody-coupled protein A/G magnetic beads were incubated with cell lysates at 4°C overnight, then the proteinase K digestion buffer was added, and RNA was extracted by TRIzol. qPCR was performed to determine the interaction between METTL1 and NET1.

Xenograft mouse model
Female BALB/c nude mice aged 4-5 weeks were assigned randomly into four groups consisting of PC9 cells (1 × 106) alone; PC9 cells with osimertinib treatment; PC9 cells + CAFs (3 × 106) with osimertinib treatment; PC9 (METTL1-KD) cells + CAFs with osimertinib treatment, and each group is comprised of six mice. Cells were subcutaneously injected into the flank of nude mice, and osimertinib administration (7.5 mg/kg/d) started 7 days later when tumors were formed. The body weights and tumor volumes were measured every week. Tumor volume = d2
× D/2, where D is the longest diameter and d is the shortest diameter. The experiments were terminated after 5 weeks, and tumor tissues were collected for further examination.

Statistical analysis
All experiments were repeated at least three times and data was presented as means ± SD. The statistical analysis was conducted using GraphPad Prism 9. The Student’s T-test was used for two-group comparisons, and one-way ANOVA was used for multi-groups comparisons. p < 0.05 was considered statistically significant.

Results

CAFs enhance osimertinib resistance in NSCLC cells
Numerous studies have revealed that CAFs are involved in chemoresistance in various types of cancer. Our previous study in lung cancer showed that CAFs enhanced cisplatin resistance [5], this finding drove us to further investigate the role of CAFs in the resistance of NSCLC cells to the targeted drug - osimertinib.
CAFs and NFs were isolated from the specimens of NSCLC patients, and EGFR-mutant NSCLC cells lines PC9 (EGFR del19) and H1975 (L858R/T790M) were chosen to examine the therapeutic effect of osimertinib. NSCLC cells were cultured in CAF-CM or NF-CM, then treated with osimertinib. Cell viability was detected by CCK-8 kit. Figure 1A–C showed that both CAFs and NFs reduced the sensitivity of NSCLC cells to osimertinib, and CAFs were more effective. The CAFs’ effect on cell growth was further determined by colony formation. CAFs enhanced colony formation of NSCLC cells not only in untreated cells, but also in osimertinib-treated cells (Fig. 1D–G).
Because osimertinib may inhibit NSCLC cell growth via inducing apoptosis [17, 18], we next investigated the effect of CAFs on apoptosis induction. As shown in Fig. 1H–K, CAFs did not affect the apoptosis level in untreated cells, however, CAFs significantly reduced the apoptosis level in osimertinib-treated cells. We then examined apoptosis-related genes, and found that CAFs suppressed caspase-3 activation, upregulated Survivin and Bcl-2 expression (Fig. 1L). As the stemness of cancer cells plays a pivotal role in drug resistance [19], we further evaluated the effect of CAFs on cell stemness. CAFs facilitated NSCLC cell stemness by upregulating stemness-related genes SOX2, OCT4 and NANOG (Fig. 1M). Taken together, these results demonstrated that CAFs enhance osimertinib resistance in NSCLC cells by reducing apoptosis and facilitating cell stemness.

CAFs increase m7G modification in NSCLC cells
Accumulating evidence has demonstrated that aberrant m7G modification is involved in tumorigenesis and tumor progression [20], however, the effect of CAFs on m7G modification in cancer cells has not been elucidated. To evaluate the role of CAFs in m7G modification in cancer cells, NSCLC cells were cultured in CAF-CM, and the m7G modification in cancer cells were detected by Dot Blot assay. Figure 2A, B showed that both NFs and CAFs elevated m7G modification level in NSCLC cells, but CAFs had a stronger effect. Based on this finding, we next performed MeRIP-sequencing (MeRIP-seq) on NSCLC cells cultured in CAF-CM to detect m7G modification. The MeRIP-seq results also indicated that CAFs increased m7G modification level in NSCLC cells, in particular, m7G modification level was increased in CDS region and 3’UTR region, whereas, m7G modification level was reduced in 5’UTR region (Fig. 2C, D).
M7G modifications are catalyzed by methyltransferases, including METTL1/WDR4 complex, RNMT/RAM complex, and WBSCR22/TRMT122 complex [21]. Given that METTL1/WDR4 complex installs m7G modifications on mRNA, and we are interested in mRNA methylation, we focused on METTL1/WDR4 complex. We first compared the expression level of METTL1 in lung cancer cells (PC9 and H1975 cells) and normal human lung bronchial epithelial cells (BEAS-2B cells). Figure 2E, F showed that lung cancer cells expressed higher level of METTL1 at both mRNA and protein level. Importantly, METTL1 was also upregulated in tumor tissues compared with paired nontumor tissues (Fig. 2G, H). Then we investigated whether CAFs affect METTL1 expression in lung cancer cells, and found that CAFs upregulated METTL1 (Fig. 2I, J). Furthermore, we assessed WDR4 level in lung cancer cells. Similar to METTL1, WDR4 expressions were higher in NSCLC cells compared to normal lung bronchial epithelial cells, and were upregulated by CAFs (Fig. 2K, L). These data demonstrated that CAFs increased m7G modification in NSCLC cells, and METTL1 and WDR4 were modulated by CAFs.

METTL1 mediates CAFs’ effect on m7G modification and osimertinib resistance of NSCLC cells
Because METTL1 serves as a methyltransferase, while WDR4 serves as a stabilizing molecule, we then focused on the role of METTL1 in CAFs’ effect on m7G modifications. METTL1 expression was manipulated by overexpression or knockdown in NSCLC cells (Fig. 3A, B), and m7G modification was detected by Dot Blot assay. As shown in Fig. 3C, METTL1 knockdown decreased m7G modification level, whereas, METTL1 overexpression increased m7G modification level. To determine whether METTL1 is responsible for CAFs’ effect on m7G modifications, METTL1 was knocked down in NSCLC cells and cells were cultured in CAF-CM. Dot Blot assay showed that CAFs’ promoting effect on m7G modifications was mitigated, suggesting METTL1 mediated CAFs’ effect on m7G modifications (Fig. 3D).
We next investigated the role of METTL1 in CAFs’ effect on osimertinib-resistance of NSCLC cells. We first examined the effect of METTL1 on cell growth under osimertinib treatment, and found that METTL1 depletion dramatically enhanced the sensitivity of NSCLC cells to osimertinib (Fig. 3E–G). Then METTL1-knockdown NSCLC cells were culture in CAF-CM and treated with osimertinib. Interestingly, METTL1 knockdown significantly attenuated CAFs’ stimulatory effect on osimertinib-resistance (Fig. 3H–J), indicating CAFs’ facilitated osimertinib-resistance of NSCLC cells via METTL1 mediated m7G modifications.
Furthermore, we assessed the role of METTL1 in colony formation and apoptosis induction of NSCLC cells treated with osimertinib. As shown in Fig. 3K–R decreased METTL1 dramatically reduced colony number and increased apoptosis rate. These findings drove us to further evaluate the clinical significance of METTL1 in lung cancer. TCGA database analysis indicated that METTL1 was highly expressed in lung cancer patients compared to healthy people, and this was further supported by analyzing paired tumor and normal adjacent tissues (Fig. 3S, T). In addition, high level of METTL1 was associated with poor survival rate (Fig. 3U). Furthermore, METTL1 level was correlated with pathological stages of lung cancer (Fig. 3V). Collectively, these data indicated that METTL1 mediated CAFs’ effect on m7G modification and osimertinib resistance of NSCLC cells, and importantly, METTL1 was involved in lung cancer progression.

CAFs elevate METTL1 and m7G modification level in NSCLC cells via HMGB1 secretion
CAFs regulate cancer cell functions via secreting proteins such growth factors and cytokines. In our previous studies, we compared differentially expressed genes in CAFs and NFs by RNA sequencing (RNA-seq), and found that HMGB1 was highly expressed in CAFs [5]. We also revealed that CAFs promotes NSCLC cell metastasis via autophagic secretion of HMGB1 [16]. These works provided rationale for the choice of HMGB1 for further investigation in present study. To clarify the potential role of HMGB1 in CAFs’ effect on m7G modification in NSCLC cells, we first compared HMGB1 levels released from cells by ELISA assay, and found the level of CAFs secreted HMGB1 was significantly higher than that of lung cancer cells secreted (Fig. 4A). Then we determined the effect of HMGB1-derived from CAFs on METTL1 expression in lung cancer cells. As shown in Fig. 4B, C, both CAF-CM and recombinant human HMGB1 dramatically increased METTL1 expression, whereas, when HMGB1 neutralizing antibody was added to CAF-CM, the enhancing effect of CAFs on METTL1 expression was attenuated. The effect of HMGB1 on m7G modification in lung cancer cells was also investigated by applying recombinant human HMGB1 and HMGB1 neutralizing antibody. Figure 4D indicated that HMGB1 mediated the effect of CAFs on m7G modification level. These results demonstrated that CAFs elevated METTL1 and m7G modification level in NSCLC cells via HMGB1 secretion.

CAFs increase m7G modification level and expression of NET1 in NSCLC cells via METTL1
Based on our above finding that CAFs elevate m7G modification level of NSCLC cells, we next investigated which gene’s m7G modification was regulated by CAFs. M7G modification affects RNA process, regulates RNA stability and translation [22], we therefore performed RNA-seq on CAFs-treated NSCLC cells, and analyzed the differentially expressed genes. RNA-seq data showed that there were 1721 upregulated genes and 2256 downregulated genes (Fig. 5A, B). We further analyzed the RNA-seq data in association with MeRIP-seq data, and found that there were 1124 hypermethylated genes, among them 630 genes were upregulated (hyper-up), and 494 genes were downregulated (hyper-down); meanwhile, there were 1340 hypomethylated genes, among them 475 genes were upregulated (hypo-up), and 865 genes were downregulated (hypo-down) (Fig. 5C). Because m7G modification may increase RNA stability and expression, we then screened the hyper-up genes specifically with the promoting role in cancer. Among these genes NET1 was chosen as one of the candidate genes.
We first analyzed the predicted m7G modification sites in NET1 on m7GHub website (http://www.rnamd.org/m7GHub2/), and found there were 11 predicted m7G modification sites in NET1 (Fig. 5D). Then we explore the possibility of NET1 regulation by METTL1 on ENCORI website (https://rnasysu.com/encori/), and found that METTL1 may bind to NET1 mRNA (Fig. 5E). Based on above analysis, we chose NET1 for further investigation. To validate the sequencing results, NSCLC cells were cultured in CAF-CM, and the m7G modification level and expression level of NET1 were detected by MeRIP-qPCR and qPCR, respectively. As shown in Fig. 5F, G and Fig. 5H, I, both m7G modification level and mRNA level of NET1 were upregulated, which were consistent with the sequencing results. In addition, the Western blot assay indicated that the protein level of NET1 was also increased by CAFs (Fig. 5J).
Given that m7G modification of NSCLC cells were catalyzed by METTL1, we investigated whether m7G modification of NET1 was also modulated by METTL1. The direct interaction between METTL1 and NET1 transcript was examined by RIP-qPCR. By applying specific anti-METTL1 antibody, we found that CAFs dramatically increased NET1 mRNA; whereas, in METTL1 depleted cells, the enrichment of NET1 mRNA was mitigated (Fig. 5K). To further confirm m7G modification of NET1 by METTL1, METTL1 was overexpressed in lung cancer cells and m7G modification of NET1 was detected by MeRIP-qPCR. Figure 5L showed that m7G modification of NET1 was significantly elevated. Meanwhile, when METTL1 was knocked down, the enhancement of m7G modification by CAFs was abrogated (Fig. 5M), suggesting METTL1 mediated m7G modification of NET1 by CAFs.
Because m7G modification affects RNA stability and translation [22], we explored whether METTL1 is responsible for this process. METTL1 expression was manipulated, and NET1 mRNA stability was assessed by applying actinomycin D. We found that METTL1 increased NET1 mRNA stability (Fig. 5N–Q). In line with this finding, METTL1 also stimulated NET1 expression at both transcription and translation level (Fig. 5R–T). Notably, when METTL1 expression was reduced, the effect of CAFs on NET1 expression was attenuated (Fig. 5U, V). Collectively, our results demonstrated that CAFs increased m7G modification level and expression of NET1 in NSCLC cells via METTL1.

NET1 is responsible for CAFs’ stimulation on osimertinib resistance of NSCLC cells via AKT/NF-κB pathway
Next, we explored the role of NET1 in CAFs’ effect on Osimertinib resistance of NSCLC cells. Since CAFs upregulated NET1 expression, we reduced NET1 expression by RNAi in lung cancer cells, and found that NET1 depletion increased the sensitivity of NSCLC cells to osimertinib (Fig. 6A–C). We further knocked down or overexpressed NET1 in NSCLC cells, then cultured cells in CAF-CM and treated cells with osimertinib. Figure 6D–G showed that NET1 depletion attenuated, whereas NET1 overexpression stimulated the enhancement of CAFs on osimertinib-resistance of NSCLC cells. Based on above findings, we evaluated the clinical implication of NET1 in lung cancer. By analyzing TCGA database, we found that NET1 was upregulated in both total and paired lung cancer tissues, and high NET1 level was associated with poor prognosis. Furthermore, NET1 level was correlated with lung cancer stages (Fig. 6H–K).
To elucidate the signaling pathways underlying CAFs’ effect on NSCLC cells, we performed KEGG pathway analysis on our RNA-seq data. Figure 6L indicated that PI3K-AKT pathway was activated by CAFs. Further cell experiment showed that CAFs elevated p-AKT and p-p65 levels, meanwhile, total AKT and p65 levels remained unchanged (Fig. 6M). Since one of the resistance mechanisms to osimertinib is the signaling pathway activation, including PI3K-AKT pathway [2], we investigated whether AKT/NF-κB pathway is involved in CAFs’ effect on osimertinib-resistance of NSCLC cells. Lung cancer cells were pretreated with selective AKT inhibitor perifosine or NF-κB inhibitor JSH23. As shown in Fig. 6N and O, the suppression of AKT/NF-κB pathway dramatically mitigated CAFs’ enhancement on osimertinib-resistance of NSCLC cells. Moreover, we manipulated the expression of NET1 in NSCLC cells, and detected the AKT/NF-κB pathway, and found that NET1 mediated CAFs’ effect on AKT/NF-κB pathway (Fig. 6P). Taken together, these data demonstrated that NET1 is responsible for CAFs’ stimulation on osimertinib-resistance of NSCLC cells via AKT/NF-κB pathway.

Inhibition of m7G modification diminishes CAFs’ effect on lung cancer growth in vivo
To evaluate the role of m7G modification in lung cancer cell growth in vivo, we established a stable METTL1-knockdown lung cancer cell line by using lentivirus. The efficiency of lentivirus infection and METTL1 silencing were examined by observing GFP fluorescence and Western blot assay (Fig. 7A, B). Interestingly, we found that METTL1 knockdown inhibited lung cancer cell growth (Fig. 7C, D).
Next, we try to clarify the effect of m7G modification on lung cancer cell growth in vivo. Mice were divided into 4 groups, including PC9 cells, PC9 cells + osimertinib treatment, PC9 cells + CAFs + osimertinib treatment, and PC9-METTL1-KD cells + CAFs + osimertinib treatment. Tumor growth was determined by tumor volume. As shown in Fig. 7E–I, while CAFs stimulated tumor growth dramatically, METTL1-KD significantly mitigated CAFs’ stimulatory effect. Then we examined the m7G modification in tumor tissues. The Dot Blot assay indicated that CAFs increased m7G modification level, however, in PC9- METTL1-KD cells, the increasement of m7G modification by CAFs was abrogated (Fig. 7J, K).
Moreover, we explored the underlying mechanism by examining tumor tissues. Immunohistochemistry assay showed that CAFs upregulated METTL1 and NET1 expression, and METTL1-KD downregulated NET1 expression (Fig. 7L). Western blot assay showed that CAFs elevated METTL1 and NET1 expression and activated AKT/NF-κB pathway (Fig. 7M). These results were consistent with our in vitro findings, suggesting CAFs facilitated tumor growth through METTL1-mediated m7G modification of NET1 and activation of AKT/NF-κB pathway.

Discussion
In this study we investigated the role of CAFs in the resistance of NSCLC cells to osimertinib. We have revealed a novel mechanism that CAFs conferred osimertinib resistance in NSCLC cells through modulating m7G modification. Further study demonstrated that CAFs upregulated METTL1 in NSCLC cells by releasing HMGB1, and METTL1 elevated m7G modification of NSCLC cells. By applying MeRIP-seq and RNA-seq, we identified NET1 as a target of METTL1, and enhanced m7G modification of NET1 increased NET1 expression and activated downstream AKT/NF-κB pathway. Importantly, reducing m7G modification by METTL1 knockdown significantly attenuated CAFs’ stimulatory effect on osimertinib resistance both in vitro and in vivo (Fig. 8).
For many decades, cancer research has been focused on cancer cells alone, until recently people realize the communication between cancer cells and TME plays a crucial role in cancer development. As one of the major components of TME, CAFs contribute to drug resistance in various types of cancer. Interestingly, recent studies have revealed that CAFs are also responsible for osimertinib resistance in lung cancer. CAFs promote osimertinib resistance in lung adenocarcinoma by enhancing ribosome biosynthesis via releasing colony-stimulating factor 2 (CSF2) [9]. CAFs-derived periostin facilitates osimertinib resistance in NSCLC cells by activating ERK signaling pathway and inducing EMT [10]. Notably, a clinical study indicates that high podoplanin level in CAFs may predict primary resistance to osimertinib in lung adenocarcinoma [23]. However, whether m7G modification is responsible for CAFs’ effect on drug resistance has not been evaluated. In this study, we have revealed a novel mechanism that CAFs enhanced osimertinib resistance in NSCLC cells by modulating METTL1-mediated m7G modification of NSCLC cells.
Accumulating evidence indicates that m7G modification play a crucial role in cancer drug resistance, and most studies are focused on m7G modification on tRNA. METTL1/WDR4-mediated m7G tRNA modification stimulates the translation of EGFR pathway genes and triggers lenvatinib resistance in hepatocellular carcinoma [24]. METTL1-mediated m7G tRNA modification activates the WNT/β-catenin signaling pathway and promotes chemoresistance to cisplatin and docetaxel in nasopharyngeal carcinoma [25]. M7G tRNA modification may also drive a metabolic shift from glycolysis to oxidative phosphorylation to enhance the acquired anlotinib resistance in oral squamous cell carcinoma [26]. However, there are very limited studies working on m7G modification on mRNA in cancer drug resistance. He M et al. reported that m7G modification of FTH1 mRNA and pri-miR-26a regulated chemoresistance to doxorubicin and platinum in osteosarcoma. Mechanistic study revealed that METTL1-mediated m7G modification of FTH1 and pri-miR-26a results in reduced ferroptosis and leads to chemoresistance [27]. Interestingly, we found that CAFs regulated m7G modification of NET1 mRNA and activated downstream AKT/NF-κB pathway to facilitate osimertinib resistance. Taken together, these studies indicate that m7G modifications contribute to drug resistance via different mechanisms.
Several studies indicate that HMGB1 plays an important role in cancer drug resistance. EZH2-mediated PHF10 suppression activates HMGB1/NF-κB axis to stimulate gemcitabine resistance in cholangiocarcinoma [28]. SIRT1 depletion induces ferroptosis and mitigates cytarabine resistance in acute myeloid leukemia through HMGB1/ACSL4 pathway [29]. Gfi-1 inhibits HMGB1 to reduce autophagy level of colorectal cancer cell and overcome oxaliplatin resistance [30]. Of note, in lung cancer, HMGB1 is highly expressed in pleural effusion and recurrent tissues of NSCLC. Overexpression of HMGB1 in NSCLC cells contributes to cisplatin resistance both in vitro and in vivo [31]. Lei et al. reported that HMGB1 expression level is higher in gefitinib resistant NSCLC cells than in sensitive cells. Mechanistic study revealed that high HMGB1 level promotes gefitinib resistance by inducing autophagy [32]. However, these studies were focused on the role of HMGB1 in NSCLC cells, the role of CAFs-derived HMGB1 has not been evaluated. Our study revealed that HMGB1 secreted by CAFs upregulated METTL1 in NSCLC cells and elevated m7G modification, which contributed to osimertinib resistance. All these studies demonstrated that HMGB1 plays a pivotal role in drug resistance via diverse mechanisms, suggesting HMGB1 as a potential therapeutic target for drug resistance.
NET1 is a member of Rho guanine nucleotide exchange factors family, it is involved in various types of cancer. NET1 is highly expressed in colon cancer cells, it boosts colon cancer cell proliferation and suppresses apoptosis [33]. In acute lymphoblastic leukemia (ALL), circ_0000745 upregulate NET1 expression to promote ALL progression [34]. Zhao et al. reported that baicalin increased miR-340-5p expression whereas reduced miR-340-5p target NET1 expression, and suppressed lung cancer cell growth [35]. Although the role of NET1 has been investigated in many cancer types, the m7G modification of NET1 has not been studied. Here, we found that CAFs increased m7G modification and expression of NET1 via METTL1. One of the mechanisms underlying osimertinib resistance is signaling pathway activation, including RAS-MAPK pathway, JAK-STAT pathway, and PI3K-AKT pathway [2]. Interestingly, the KEGG pathway analysis on our RNA-seq data indicated that AKT pathway is activated by CAFs, and this drove us to further explore the role of AKT signaling pathway in CAFs’ promoting effect on osimertinib resistance, and how AKT pathway was activated by CAFs. We found the CAFs activated AKT signaling pathway by upregulating NET1, and AKT pathway activation is responsible for CAFs’ stimulation on osimertinib resistance of NSCLC cells. Importantly, by analyzing the TCGA database, we found that both NET1 and METTL1 were highly expressed in NSCLC patients, associated with poor prognosis, and correlated with pathological stages of lung cancer. Further investigations are needed to evaluate whether NET1 and METTL1 could serve as potential biomarkers for predicting response to osimertinib in NSCLC patients.
The limitation of this study should be noted. Recent studies have demonstrated the heterogeneity of CAFs, and several subtypes of CAFs have been identified, including myofibroblastic CAFs (myCAFs), inflammatory CAFs (iCAFs), and antigen-presenting CAFs (apCAFs) [36, 37]. Notably, in lung cancer, some new CAFs subtypes have been found with specific functions. Hu et al. identified three subtypes of CAFs with robustly protective role, moderately protective role, and minimal protective role of cancers [38]. Kim et al. classified CAFs into “immunosuppressive”, “neoantigen presenting”, “myofibroblastic”, and “proliferative” subtypes [39]. Importantly, Cords et al. established a prognostic model for NSCLC based on 11 CAF phenotypes [40]. In our study, the CAFs were isolated from lung cancer specimens, however, these CAFs were a mixture of CAFs subpopulation. In order to further elucidate the functional heterogeneity of lung cancer CAFs, and to identify the specific subtypes that contribute to osimertinib resistance, an in-depth investigation is urgently needed. There were also limitations in our in vivo model. The mice we used were immunodeficient mice, which were lack of immune components. Lung cancer cells were subcutaneously injected into the flank of nude mice, instead, the orthotopic model will be idea for a lung cancer model. Furthermore, the PDX model will be better than single cancer cell line model.
In summary, this study revealed that CAFs promoted the resistance of NSCLC cells to osimertinib via m7G modification-dependent mechanism, and METTL1 mediated the m7G modification regulation. Further study demonstrated that METTL1 elevated m7G modification level and expression of NET1, and NET1 activated AKT/NF-κB pathway to facilitate osimertinib resistance. These findings underscore the importance of m7G modification in the communication between cancer cells and the TME, suggesting that targeting the CAF-METTL1-NET1 m7G modification-AKT/NF-κB axis may represent a novel strategy to overcome osimertinib resistance.

Supplementary information